On Thu, Oct 30, 2003 at 04:36:02PM +0000, Ian Clarke wrote:
> There are a few significant things remaining before NGR can really be 
> considered "fully implemented" as-per the original vision.
> 
> Mainly this involves allowing nodes to pass around the NGR data they 
> have collected embedded in node references, such that when a node learns 
> about a new node in the network, it gets a head-start when it comes to 
> probing its performance and capabilities.
> 
> There are two main areas in which this will require changes:
> 
> 1) The DataSource in DataReply messages.
> 
> 2) The seednodes.ref mechanism.
> 

4) The Secondary Failure Table. Which will help with both stats and
accuracy. Well OK it wasn't in the original vision.
5) The diagnostic Ian proposed - the difference between the estimated
time and the actual time. Of course this will be a bit tricky to get
right, as most requests fail, and I personally have always held that it
will have ludicrous error bars, but over a long period it could be
interesting - it measures whether NGRouting works on its own terms.

> And potentially:
> 
> 3) The announcement mechanism

What do you want to do with this? The announce key is used to create
estimators initially specialized in it; this is already implemented.
> 
> In (1) it is important to have safeguards against abuse of this 
> mechanism, such as a node giving itself an extremely flattering estimate 
> to encourage traffic.  One way to do this is to modify the behavior of 
> any node passing back a DataReply with a DataSource for a node with 
> which it has communicated to replace the estimator with its own.

Also, we should not serialize the point age data - it should all start 
at zero.
> 
> Thoughts?
> 
> Ian.

-- 
Matthew J Toseland - [EMAIL PROTECTED]
Freenet Project Official Codemonkey - http://freenetproject.org/
ICTHUS - Nothing is impossible. Our Boss says so.

Attachment: signature.asc
Description: Digital signature

_______________________________________________
Devl mailing list
[EMAIL PROTECTED]
http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/devl

Reply via email to